Ten models of NEX-GDDP CMIP5 were used to perform equal-weighted averaging under the RCP4.5 and RCP8.5 scenarios to obtain daily precipitation and temperature data under a multi-model ensemble. The CREST and VIC models were used to project the change characteristics of runoff and precipitation in the Xijiang River Basin under the background of a global warming by 1.5 °C and 2 °C, respectively. The results show that: (1) under the two warming target scenarios, there are obvious regional differences in the extreme precipitation in the Xijiang River Basin under the RCP4.5 and RCP8.5 scenarios. The precipitation increases on the whole and more so under the high-emission and greater-warming scenarios. In addition, extreme precipitation events in the Xijiang River Basin are significantly different at a temperature rise of additional 0.5 °C. (2) CREST and VIC have good feasibility in the Xijiang River Basin. The projected runoff increases under different combinations of scenarios and at various time scales compared to the baseline period. (3) There is no significant difference between the multi-annual average monthly runoff distribution percentage calculated by the multi-model and hydrological model ensemble average and the multi-annual average monthly runoff distribution percentage during the baseline period and the distributions under the RCP 4.5 and 8.5 scenarios are more uniform and uneven, respectively, than that in the baseline period.